For years, the energy industry was preparing for a future defined by efficiency. Cars would consume less fuel. Homes would use less power. Renewable generation would lower costs. Demand growth would slow. The biggest debate wasn’t whether America would have enough electricity — it was what kind of electricity it would use.
Then artificial intelligence arrived.
What began as a technology story is rapidly becoming one of the largest infrastructure stories in modern American history. Utilities are rewriting demand forecasts. Grid operators are warning about supply shortages. Natural gas developers are accelerating projects. Communities are organizing against data centers. Politicians are discovering that the race for AI dominance comes with a cost measured not in processors or software, but in megawatts.
The market remains obsessed with the companies building AI models. The more interesting question may be whether the infrastructure needed to power those models can be built quickly enough to support them.
The most important energy development of the week wasn’t a merger, an acquisition, or a billion-dollar LNG project. It was a regulatory admission.
The Federal Energy Regulatory Commission approved PJM Interconnection’s plan to fast-track major power projects in response to surging electricity demand. PJM operates the largest power market in North America, serving roughly 67 million people across 13 states, and the organization has become increasingly vocal about the challenge posed by data center growth and electrification. Under the new framework, select projects can move through the interconnection process significantly faster than under traditional timelines, an acknowledgment that the existing system is struggling to keep pace with demand. [1]
That matters because interconnection queues have quietly become one of the largest bottlenecks in American infrastructure. For years, the conversation around energy focused on whether enough generation would be built. Increasingly, the problem is whether generation can connect to the grid before demand arrives.
The implications extend far beyond utilities. Every new AI model requires more computing power. More computing power requires more servers. More servers require more data centers. Those data centers require transmission infrastructure, substations, pipelines, transformers, backup generation, and reliable baseload electricity.
The technology sector may be creating the demand, but it is the infrastructure sector that will be asked to meet it.
That distinction matters because infrastructure booms tend to create value in places investors initially overlook. During a gold rush, the most reliable profits often belong to the companies selling shovels.
While regulators were trying to accelerate new generation, Reuters published one of the most revealing investigations of the year.
Across the United States, dozens of natural gas power plants are being rapidly developed to serve AI data centers. Many are moving through approval processes in a fraction of the time traditionally associated with major energy projects. In some cases, residents only learned about nearby developments after construction had already begun. Researchers identified at least 57 off-grid power plants proposed or under development to support individual data centers, representing roughly 73,000 megawatts of generating capacity. [2]
That figure should stop people in their tracks.
For years, conventional wisdom held that renewables would steadily replace fossil fuel generation. Yet the first major infrastructure response to the AI boom has not been solar farms or wind projects. It has been natural gas.
The reason is simple: reliability.
AI facilities do not operate on political timelines. They do not wait for future transmission upgrades, future battery installations, or future permitting reform. They require enormous amounts of electricity today. When reliability becomes the priority, natural gas remains one of the fastest scalable sources of dispatchable power available.
This does not mean the energy transition has failed. It means the transition has collided with reality.
The market increasingly appears to be moving toward an uncomfortable conclusion: the future may be renewable, but the bridge to get there is likely to be far more dependent on natural gas than many policymakers expected.
The companies building that bridge may ultimately benefit more than those debating how long it should be.
The biggest risk facing the AI buildout may not be electricity generation.
It may be public opinion.
A Reuters/Ipsos poll released this week found that only one-third of Americans support the current pace of data center construction. Nearly two-thirds oppose it, while 77% expressed concern that AI expansion could increase electricity costs. More than half said they would oppose a data center being built in their own community. [3]
Those numbers are remarkable because they cut across traditional political divisions. Concerns about energy costs, land use, water consumption, and infrastructure strain are emerging among Republicans, Democrats, and independents alike.
The political implications are significant. Support for AI development remains high at the federal level, where policymakers increasingly frame artificial intelligence as a strategic competition with China. Local communities, however, are beginning to view projects through a different lens.
Who pays for the new substations?
Who pays for the transmission lines?
Who pays when grid upgrades increase electricity costs?
More importantly, who benefits?
The backlash is already producing tangible results. More than 75 data center developments worth an estimated $130 billion have reportedly faced delays, cancellations, or successful opposition campaigns during the first months of 2026. Communities across the country are increasingly challenging projects over concerns about power demand, water usage, noise, environmental impacts, and tax incentives. [4][5]
Investors often focus on technological risk. Political risk is frequently underestimated until it arrives.
The next major battle in AI may not be fought over semiconductors or software models. It may be fought in city council chambers, public hearings, utility commissions, and state legislatures.
Because once voters believe they are subsidizing AI infrastructure through higher electricity bills, the conversation stops being about technology and starts becoming about affordability.
And affordability has a way of reshaping markets very quickly.
The energy industry spent years debating oil versus renewables, fossil fuels versus clean energy, and growth versus decarbonization.
The AI boom is forcing a different conversation.
Can America build enough infrastructure fast enough to support its own ambitions?
The answer increasingly appears uncertain.
Grid operators are accelerating approvals. Utilities are revising forecasts. Natural gas developers are expanding projects. Communities are pushing back. Politicians are trying to balance economic growth against rising public concerns.
None of this looks like a traditional energy crisis.
But it may be the early stages of one.
Not because America is running out of energy, but because demand is arriving faster than the infrastructure required to support it.
For investors, that distinction could prove extremely important over the next decade.
The market may continue focusing on the companies creating artificial intelligence.
The larger opportunity may be found in the companies building the physical systems that make it possible.
[1] https://www.reuters.com/legal/litigation/us-energy-regulator-approves-pjms-fast-tracked-power-plant-interconnection-plan-2026-06-10/
[2] https://www.reuters.com/business/energy/fast-tracked-power-plants-fuel-ai-boom-with-little-public-scrutiny-2026-06-16/
[3] https://www.reuters.com/world/us/americans-wary-ai-driven-data-center-boom-reutersipsos-poll-shows-2026-06-11/
[4] https://www.tomshardware.com/tech-industry/artificial-intelligence/more-than-75-data-center-build-outs-worth-usd130-billion-have-been-successfully-blocked-in-the-first-four-months-of-2026-bipartisan-opposition-mounts-nationwide-over-fears-of-soaring-power-and-water-costs
[5] https://www.businessinsider.com/data-center-bans-moratoriums-opposition-map-2026-6
Your personal details are strictly for our use, and you can unsubscribe at any time.